摘要
工业生产中螺母的漏装会导致严重的安全隐患,因此螺母的识别与检测有着重要意义。螺母可提取特征单一,使用传统方法检测时对光照、噪声等变化极为敏感,准确率较低。研究提出一种基于卷积神经网络的螺母检测算法,利用单目标多窗口检测器(single shot multibox detector,SSD)快速检测图像中可能的目标窗口,使用AlexNet对这些窗口进行分类以获得更高的分类精确度,使用基于面积比的非极大值抑制算法去除重复窗口得到最终检测结果。实验表明,该算法检测图像速度快且准确率高,可对螺母漏装进行高精度检测。
In industrial production,nut missing can lead to serious safety hazards,so nut identification and detection is of great significance.Since the extraction features of the nut are relatively simple and the traditional detection method is sensitive to the changes of illumination and noise,the detection accuracy is very low.A nut detection algorithm based on convolutional neural network is proposed in this study.Firstly,a single shot multibox detector is used to quickly detect the possible target windows in the image.Then,these windows are classified by AlexNet to achieve higher classification accuracy.Finally,the non-maximal suppression algorithm based on the area ratio is utilized to remove the duplicate window to generate the detection results.The results indicate that the algorithm can detect the image with fast speed and high accuracy,which can realize high precision detection on the missing nut.
作者
苏思悦
付莹
来林静
SU Siyue;FU Ying;LAI Linjing(School of Computer Science,Beijing Institute of Technology,Beijing 100081,China)
出处
《中国科技论文》
CAS
北大核心
2018年第4期414-419,共6页
China Sciencepaper
关键词
计算机视觉技术
螺母检测
螺母分类
卷积神经网络
computer vision technology
nut detection
nut classification
convolutional neural network
